Diagnostic Microbiology and Infectious Disease 94 (2019) 255–259

Contents lists available at ScienceDirect

Diagnostic Microbiology and Infectious Disease

journal homepage: www.elsevier.com/locate/diagmicrobio

Clinical Studies Microbiology of polymicrobial prosthetic joint

Laure Flurin a, Kerryl E. Greenwood-Quaintance a, Robin Patel a,b,⁎ a Divisions of Clinical Microbiology, Mayo Clinic, Rochester, MN b Infectious Diseases, Mayo Clinic, Rochester, MN article info abstract

Article history: Prosthetic joint infection (PJI) is a rare but challenging complication of arthroplasty. Herein, we describe the Received 25 May 2018 epidemiology and microbiology of PJI, with a focus on analyzing differences between the microbiology of Received in revised form 31 December 2018 polymicrobial versus monomicrobial infection of hip, knee, and shoulder prostheses. In addition, we report the Accepted 7 January 2019 most frequent co- in polymicrobial , as detected by culture. A total of 373 patients diagnosed Available online 15 January 2019 with PJI at Mayo Clinic were studied. For hip and knee arthroplasties, a higher proportion of fractures (P =0.02) and a shorter time between the implantation and symptom onset (P b 0.0001) were noted in polymicrobial versus monomicrobial PJI. The most common microorganism detected, Staphylococcus epidermidis, was more frequently detected in polymicrobial (60%) versus monomicrobial (35%) PJI (P = 0.0067). Among polymicrobial infections, no co-pathogens were more frequently found than others, except S. epidermidis and faecalis which were found together in 5 cases. In addition to -negative staphylococci and enterococci, Corynebacterium species and magna were common in polymicrobial infections. Conversely, there was no difference between the prevalence of , Gram-negative , or Cutibacterium acnes between the polymicrobial and monomicrobial groups. The microbiology of polymicrobial PJI is different from that of monomicrobial PJI. © 2019 Elsevier Inc. All rights reserved.

1. Introduction anaerobic (4%), and enterococci (3%) (Tande and Patel, 2014). For shoulder arthroplasties, the microbiology is different, as the most Joint replacement is a common surgical procedure, improving frequent microorganisms are Cutibacterium acnes, followed by symptoms of degenerative joint disease by relieving pain and increasing Staphylococcus epidermidis and S. aureus. joint mobility (Osmon et al., 2013). Indeed, the prevalence of hip and Polymicrobial PJI is conversely found in ~15% of the cases and is not knee arthroplasty is 0.8% and 1.5%, respectively, of the United States well understood; for example, it is not mentioned in the Infectious population, and the number of shoulder replacements, though smaller, Disease Society of America (IDSA) guidelines on PJI (Osmon et al., has been increasing over the last 2 decades (Day et al., 2010; Maradit 2013). Moreover, polymicrobial PJIs appear to have a different microbi- Kremers et al., 2015). ology than their monomicrobial counterparts. But there has been little Although arthroplasties have high success rates, 1.5% to 2.5% are research precisely describing polymicrobial PJI, mostly because of the complicated by prosthetic joint infection (PJI), leading to high medical relatively small numbers of cases and the historical limited identifica- costs and complex patient care (Kurtz et al., 2012; Lentino, 2003). tion of associated bacteria, a situation that has improved with the Because the bacteria involved grow as biofilms on both the prosthe- advent of matrix-assisted laser desorption ionization time-of-flight sis and in the surrounding native tissue, they can be difficult to mass spectrometry (MALDI TOF MS) (Tande and Patel, 2014). As a re- eradicate and, as a result, often require long antibiotic therapy and sult, little information is available as to the frequency of the different additional surgeries. bacteria involved in polymicrobial PJI and whether certain microorgan- In order to adapt empirical antibiotic therapy before results of cul- isms have more of a predilection than others to be found in such cases. tures, an accurate understanding of the microbiology of PJI is needed. Biofilms are structured microbial entities on surfaces consisting of It is estimated that ~85% of PJIs involve a single bacterium. For hip and clusters of bacteria surrounded by a polymeric matrix (Zimmerli et al., knee arthroplasties, the 2 most common microorganisms found are 2004). Although biofilms are often studied as monomicrobial rather coagulase-negative Staphylococcus species (27%) and Staphylococcus than polymicrobial structures, they may be either. A recent trend in bac- aureus (27%), followed by Gram-negative bacilli (9%), streptococci (8%), teriology is to study polymicrobial biofilm formation in vitro and in vivo. This is justified because polymicrobial biofilms are found at significant rates in infections such as those occurring in wounds (Elias and Banin, ☆ Presented in part at ECCMID, Madrid, Spain, April 21–24, 2018 2012; Gabrilska and Rumbaugh, 2015); however, this concept may ⁎ Corresponding author at: Division of Clinical Microbiology, Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN 55905. also be applicable to PJI. Certain types of microorganisms may be more E-mail address: [email protected] (R. Patel). likely to grow together than separately, and there may be benefits in https://doi.org/10.1016/j.diagmicrobio.2019.01.006 0732-8893/© 2019 Elsevier Inc. All rights reserved. 256 L. Flurin et al. / Diagnostic Microbiology and Infectious Disease 94 (2019) 255–259 doing so (Tyner and Patel, 2016). For example, we showed that et al., 2007; Vergidis et al., 2011). We quantified and identified the dual-species S. aureus and C. acnes biofilms enhanced survival of sonicate fluid isolates using standard laboratory techniques, including S. aureus (Tyner and Patel, 2016). This led us to hypothesize that some MALDI TOF MS. For sonicate fluids collected from August 2003 to species may be more likely than others to form polymicrobial biofilms December 2005, we considered a culture significant if there was growth on prosthetic joints. of greater than 5 colony-forming units (cfu)/0.5 mL of sonicate fluid To test this hypothesis as well as to explicitly describe the micro- (Trampuz et al., 2007). For those collected from January 2006 to biology of polymicrobial PJI, we conducted a retrospective epidemi- March 2017, because we used a concentration step, we considered ologic study in which we analyzed 373 positive sonicate fluid cultures significant if there was growth of greater than 20 cfu/10 mL cultures from 373 patients with infected hip, knee, or shoulder of sonicate fluid (Cazanave et al., 2013; Piper et al., 2009). arthroplasties. We then compared the microbiology of polymicrobial versus monomicrobial infections for the 3 types of joint replacements. 2.4. Definitions

2. Materials and methods We considered a PJI to be monomicrobial if only 1 bacterial species had grown at a significant level from sonicate fluid cultures. In contrast, 2.1. Study design and population we considered a PJI polymicrobial if more than 1 species was isolated at asignificant level from sonicate fluid cultures. We studied 373 subjects whose specimens had been archived in the Mayo Clinic Prosthetic Joint Infection Biobank (Rochester, MN) and 2.5. Ethics reviewed their clinical and microbiological findings. In total, we retro- spectively reviewed 373 positive sonicate fluid cultures from patients Samples were collected under the Mayo Clinic Institutional Review diagnosed with PJI who underwent surgery at the Mayo Clinic between Board protocol 09-000-808. August of 2003 and March of 2017. Subjects were included provided they had a diagnosis of a PJI and significant positive sonicate culture 2.6. Statistical analysis results. If patients had more than 1 removed implant, only the first implant removal was included. We described baseline characteristics of the study population as fre- Patient characteristics recorded according to the affected joint (hip quencies and percentages, and compared results using χ2 tests for qual- and knee, or shoulder), included age, gender, underlying joint disorder itative variables and t tests for continuous variables. For shoulder PJI, that led to the first arthroplasty, time elapsed between the first implant because the population size of the polymicrobial group was less than and infection-associated symptom onset, and whether a communicat- 5, we used a Wilcoxon test for continuous variables and a Fisher test ing sinus tract was present at the time of diagnosis were analyzed. The for qualitative variables. time to onset after implant surgery was defined as the number of days We summarized microorganism frequencies in both monomicrobial between the implant surgery and onset of symptoms. For patients and polymicrobial infection as frequencies and percentages and com- who had chronic joint pain or limited range of motion beginning a pared results using χ2 tests when the value was greater than 5. When few days after surgery, we defined the time to onset as the number of fewer than 5 variables where analyzed, we used a Fisher exact text. days between the implant surgery and the date of the diagnosis. P values b0.05 were considered significant.

2.2. Diagnosis of PJI 3. Results

Based on the IDSA guidelines, a patient was diagnosed with PJI if 1 of A total of 373 subjects with PJI met inclusion criteria and were the following criteria was met: presence of intraoperative purulence, analyzed. Because the microbiology of joint infection differs according acute inflammation on periprosthetic tissue histopathology, sinus tract to location, we separated PJIs by the joint type affected; 310 hip and communicating with the prosthesis, or positive microbiology based on knee arthroplasties were combined, and 63 shoulder arthroplasties sonicate fluid culture (Osmon et al., 2013). were separately analyzed. We then separated them into 2 groups: polymicrobial infections and monomicrobial infections (Fig. 1). 2.3. Sonicate fluid cultures 3.1. Patient characteristics Prosthetic components removed from patients who had surgery for PJI were collected in a solid sterile container and subjected Patient characteristics, separated by monomicrobial and polymicrobial to vortexing/sonication culture, as previously described (Trampuz PJI, are shown in Table 1.

373 Subjects with PJI and significant sonicate fluid cultures

310 Hip or knee 63 Shoulder arthroplasty arthroplasty

laiborcimonoM872 32 Polymicrobial aiborcimonoM06 l 3 Polymicrobial

Fig. 1. Flowchart showing the distribution of monomicrobial and polymicrobial PJI cases according to hip or knee versus shoulder arthroplasty. L. Flurin et al. / Diagnostic Microbiology and Infectious Disease 94 (2019) 255–259 257

Table 1 Patient characteristics.

Hip or knee arthroplasties Shoulder arthroplasties

Monomicrobial Polymicrobial P value Monomicrobial Polymicrobial P value

Total 278 (90%) 32 (10%) 60 (95%) 3 (5%) Sex: male 161 (58%) 14 (44%) 0.12 44 (73%) 3 (100%) 0.56 Age, mean in years 66 (IQR 58–75) 65 (IQR 58–71) 0.8 63 (IQR 57–65) 56 (IQR 52–70) 0.85

Underlying joint disorder Osteoarthritis 157 (57%) 14 (44%) 0.17 35 (58%) 2 (67%) 0.77 Fracture 30 (11%) 8 (25%) 0.02 14 (23%) 1 (33%) 0.69

Clinical presentation Sinus tract 78 (28%) 14 (44%) 0.065 9 (15%) 1 (33%) 0.41 Time of onset, mean ± SD in years 1452 ± 4231 193 ± 295 b0.0001 177 ± 1359 306 ± 530 0.26

IQR = interquartile range.

For hip and knee arthroplasties combined, the proportion of in polymicrobial PJI, which was greater than in monomicrobial infec- polymicrobial infections was 10%, and there was no difference in tions (P = 0.0002). In addition, enterococci were more frequent age, gender, osteoarthritis, or the presence of a sinus tract between in polymicrobial than monomicrobial hip and knee PJIs, reaching a the monomicrobial and polymicrobial groups. For the monomicrobial prevalence of 28% (P b 0.0001). Corynebacterium species were identified and polymicrobial hip and knee PJI groups, the median ages were 66 in 5 infections (16%) and were more common in polymicrobial than and 65 years, respectively. For hip and knee arthroplasties combined, monomicrobial hip and knee PJIs (P = 0.0007). Among the anaerobic the proportion of fractures was higher in the polymicrobial than in bacteria, F. magna was detected in 19% of polymicrobial hip and knee the monomicrobial group (P = 0.02). The time to onset of first symp- PJIs, which was more frequent than in monomicrobial hip and knee toms of hip and knee PJI combined was also shorter in polymicrobial PJIs (P b 0.001). There was no difference in prevalence of C. acnes PJI, with a mean of 193 days versus 1452 days with monomicrobial PJI between monomicrobial (3%) and polymicrobial (9%) hip and knee (P b 0.0001). PJIs (P = 0.1). Other microorganisms were detected in12% of There were just 3 polymicrobial shoulder PJIs, a rate that was polymicrobial hip and knee PJIs. lower than that of hip and knees PJIs: 5%. There were no significant Lastly, there was no difference between monomicrobial and differences in patient characteristics between the monomicrobial polymicrobial infections for Gram-negative bacilli (P = 0.7) or strepto- and polymicrobial shoulder PJI groups; however, there were only 3 cocci (P = 1). Among hips and knees infected by a sp., patients with polymicrobial shoulder PJI. non-viridans streptococci represented 3% of the polymicrobial infec- tions and 6% of the monomicrobial infections, while viridans strepto- 3.2. Microbiology cocci represented 3% of polymicrobial and 5% of monomicrobial PJIs. Because of the high prevalence of S. epidermidis in polymicrobial hip Monomicrobial hip and knee PJIs were mainly caused by S. epidermidis and knee PJIs, we analyzed its co-pathogens (Fig. 2c). Enterococcus spe- (n = 97, 35%) and S. aureus (n = 58, 21%), followed by Streptococcus cies (5 and 1 ) were found species (n = 31, 11%), Gram-negative bacilli (n = 21, 8%), Enterococcus with S. epidermidis in 6 subjects. S. aureus, Staphylococcus haemolyticus, species (n = 16, 6%), non-epidermidis coagulase-negative Staphylococcus and Staphylococcus lugdunensis were each identified twice with species (CoNS) (n = 17, 6%), Corynebacterium species (n = 8, 3%), S. epidermidis. Other co-pathogens that were found once each included (n =6,2%),andanaerobicGram-positivebacteria Staphylococcus pseudintermedius, Staphylococcus saccharolyticus plus (n = 13, 5%), including C. acnes (n =9,3%),Finegoldia magna (n =4,1%), Corynebacterium species, Cutibacterium acnes, Cutibacterium avidum, and others (n = 11, 4%) (Table 2; Fig. 2a). species, and Escherichia coli plus Corynebacterium aurimucosum. Polymicrobial hip and knee PJIs had a different distribution of There were 25 cases of hip or knee PJI caused by enterococci, of which species compared to their monomicrobial counterparts (Table 2). 9 were polymicrobial (36%). The co-pathogens were S. epidermidis for 6 S. epidermidis was found in 59% of 32 polymicrobial hip and knee PJIs (66%), 1 non-epidermidis CoNS, 1 S. aureus, and 1 Dermatobacter hominis. (Fig. 2b). The proportion of S. epidermidis was greater in polymicrobial In monomicrobial shoulder PJIs, the following organisms were than monomicrobial hip and knee PJIs (P =0.007)(Table 2). S. aureus detected: C. acnes (53%), S. epidermidis (20%), S. aureus (12%), F. magna remained present at a rate of 22% in polymicrobial PJI with no significant (5%), C. avidum (3%), and other species (7%) (Fig. 3). The sample size difference between monomicrobial and polymicrobial hip and knee for polymicrobial shoulder PJI was small (n = 3, 5% of infections). PJIs (P = 0.9). Non-epidermidis CoNS were observed at a rate of 25% One polymicrobial shoulder PJI involved C. acnes, S. epidermidis,and Corynebacterium accolens. The other 2 were associated with C. acnes Table 2 and S. epidermidis, and C. acnes and Bacillus species. Notably, C. acnes Microbiology of hip and knee PJI, sorted by monomicrobial and polymicrobial cases. was found in all 3 polymicrobial shoulder PJIs, and S. epidermidis was

Microorganism Monomicrobial Polymicrobial P value present in 2 of the 3. Seven bacteria were not identified to the species level Staphylococcus epidermidis 97 (35%) 19 (59%) 0.007 (2 Corynebacterium species, 2 Bacillus species, 1 CoNS, and 2 viridans Staphylococcus aureus 58 (21%) 7 (22%) 0.9 “ Other coagulase-negative 17 (6%) 8 (25%) 0.0002 streptococci). The CoNS was categorized into the co-pathogens with- Staphylococcus sp. out S. epidermidis” group to minimize bias in our main analysis (Fig. 2b). Enterococcus sp. 16 (6%) 9 (28%) b0.0001 Corynebacterium sp. 8 (3%) 5 (16%) 0.0007 4. Discussion Gram-negative bacilli 21 (8%) 3 (9%) 0.7 Streptococcus sp. 31 (11%) 2 (6%) 1 Granulicatella adiacens 6 (2%) 0 (0%) - This study is the first to systematically report the microbiology of Finegoldia magna 4 (1%) 6 (19%) b0.0001 polymicrobial PJIs. Overall, 450 of 457 microorganisms were identified Cutibacterium acnes 9 (3%) 3 (9%) 0.1 to the species level, making it possible to describe the microbiology of Others 11 (4%) 4 (12%) - polymicrobial versus monomicrobial PJI in a relatively large sample 258 L. Flurin et al. / Diagnostic Microbiology and Infectious Disease 94 (2019) 255–259

a b

Staphylococcus epidermidis Co-pathogens

7 6 5 4 3 2 1 0

c

Fig. 2. a. Microbiology of monomicrobial hip and knee PJI. LEGEND: *1 Aggregatibacter aphrophilus,1Capnocytophaga canimorsus,1Mycobacterium smegmatis,3Candida albicans, 1 Candida parapsilosis 1 Actinomyces naeslundii, 1 Actinomyces neuii, 1 Bacillus fragilis, 1 sporogenes, 1 Peptoniphilus asaccharolyticus **1 Corynebacterium afermantans,1Corynebacterium amycolatum,1Corynebatcerium jeikeium,1Corynebacterium propinquum,1Corynebacterium pyruviciproducens,3Corynebacterium striatum. ***12 Enterococcus faecalis,3Enterococcus faecium,1Enterococcus gallinarum. **** 2 Staphylococcus capitis,2Staphylococcus caprae,3Staphylococcus caprae/capitis,10Staphylococcus lugdunensis. *****2 Citrobacter koseri, 2 Escherichia coli,3Enterobacter cloacae complex, 1 Proteus mirabillis,3Serratia marcescens/ureilytica,2Haemophilus influenzae,7Pseudomonas aeruginosa. ******3 Streptococcus anginosus,6Streptococcus mitis group, 1 ,2Streptococcus salivarius,1Streptococcus sanguinis, 2 viridans group Streptococcus species, not further identified, 12 ,4Streptococcus dysgalactiae subspecies equisimilis b. Microbiology of polymicrobial hip and knee PJI. c. Staphylococcus epidermidis co-pathogens in hip and knee PJI. LEGEND: *6 co-pathogens were Enterococcus sp.: Staphylococcus epidermidis + Enterococcus faecalis (5 instances) Staphylococcus epidermidis + Enterococcus faecium(1 instance) Bacillus sp. and Corynebacterium sp. were not further identified. set. We found that polymicrobial PJI tends to occur earlier following laboratories because of the clinical focus on making a distinction arthroplasty surgery than does monomicrobial PJI. Polymicrobial PJIs between S. aureus and non-aureus staphylococci (Otto, 2009). Thus, were also associated with patients who required a prosthetic joint many studies do not provide data on specificspeciesofCoNS because of a fracture. Open fractures may serve as a portal of entry for involved in PJI, including S. epidermidis,makingitdifficult to skin microorganisms. Surprisingly, there was no statistically significant compare our results to the current literature. Still, our results are in difference in the percentage of subjects with sinus tracts between agreement with those of a Norwegian study focused on the microbi- the 2 groups, though there was a trend towards a higher proportion in ology of PJI from 278 patients, which included 27 polymicrobial cases the polymicrobial group. (10%) of which 18 (66%) were associated with CoNS (Langvatn et al., The population of microorganisms found in polymicrobial PJI 2015). Also, a study by Figa et al. compared 14 polymicrobial to was different than that found in monomicrobial PJI. S. epidermidis 24 monomicrobial PJIs caused by C. acnes,inwhichCoNSwasfound and other CoNS were particularly common in polymicrobial PJIs. in 10 of 14 polymicrobial infections, with 3 isolates identified as Historically, CoNS were rarely identified to the species level in clinical S. epidermidis (Figa et al., 2017). L. Flurin et al. / Diagnostic Microbiology and Infectious Disease 94 (2019) 255–259 259

findings are subject to recruitment bias since Mayo Clinic is a reference MICROBIOLOGY OF MONOMICROBIAL SHOULDER PJI center for PJI. Most patients come to Mayo Clinic following an interval of undiagnosed infection or after treatment failure and may have under- gone multiple prior surgical procedures, which could increase the rate Others* 7% of polymicrobial infections. We applied strict inclusion criteria to Staphylococcus classify cultures as positive to ensure that the organisms associated aureus 12% with polymicrobial infections were pathogens and not contaminants. Therefore, it is possible that some polymicrobial infections may have been excluded. In addition, it is possible that some infections were Cutibacterium mistakenly classified as monomicrobial because, when cultured, rapidly acnes growing organisms, especially when present at a majority concentra- 53% Staphylococcus epidermidis tion, may have outcompeted more slowly growing organisms. 20% To conclude, we describe in detail the microbiology of polymicrobial PJI in our patient population. The proportion with underlying fractures was higher in polymicrobial compared with monomicrobial infections, and polymicrobial PJIs occurred closer in time to the first implant Finegoldia surgery than did monomicrobial infections. We found differences Cutibacterium magna avidum 5% in the distribution of bacterial species between monomicrobial and 3% polymicrobial PJIs, as well as between different joint types. Some microorganisms, such as S. epidermidis, the most prevalent organism detected overall, were found at a higher rate in polymicrobial than monomicrobial infections. The most common co- with S. epidermidis was E. faecalis. These results can be used to inform further research into polymicrobial biofilm formation, especially with Fig. 3. Microbiology of monomicrobial shoulder PJI. LEGEND: *1 Desulfivibrio sp., 1 Enterococcus faecalis,1Pseudomonas aeruginosa,1Streptococcus agalactiae. S. epidermidis.

There are a several potential explanations of the prevalence of References S. epidermidis in polymicrobial PJI. First, S. epidermidis is a commensal bac- Cazanave C, Greenwood-Quaintance KE, Hanssen AD, Karau MJ, Schmidt SM, Gomez terium, which could be inoculated at the time of surgery, causing early- Urena EO, et al. Rapid molecular microbiologic diagnosis of prosthetic joint infection. onset infection. Stewart et al. studied S. epidermidis in vitro in a multispe- J Clin Microbiol 2013;51(7):2280–7. cies biofilm with S. aureus (Stewart et al., 2017), demonstrating that Day JS, Lau E, Ong KL, Williams GR, Ramsey ML, Kurtz SM. Prevalence and projections of fi total shoulder and elbow arthroplasty in the United States to 2015. J Shoulder Elbow S. epidermidis was the dominant species in the polymicrobial bio lm Surg 2010;19(8):1115–20. when given a lead time of 4 to 6 h, as well as under environmental condi- Elias S, Banin E. Multi-species biofilms: living with friendly neighbors. FEMS Microbiol tions of low pH or low concentrations of vancomycin (Stewart et al., Rev 2012;36(5):990–1004. fi Figa R, Muñetón D, Gómez L, Matamala A, Lung M, Cuchi E, et al. Periprosthetic joint in- 2017). On the other hand, S. aureus bio lm growth was faster under un- fection by Propionibacterium acnes: clinical differences between monomicrobial ver- stressed conditions (37 °C; pH 7) at which it overtook S. epidermidis sus polymicrobial infection. Anaerobe 2017;44:143–9. growth (Stewart et al., 2017). This could explain why S. epidermidis,a Gabrilska RA, Rumbaugh KP. Biofilm models of polymicrobial infection. Future Microbiol 2015;10(12):1997–2015. slow-growing bacterial species, was found in 60% of polymicrobial PJIs. Kurtz SM, Lau E, Watson H, Schmier JK, Parvizi J. Economic burden of periprosthetic joint Under stressful conditions, like in prosthetic joints, S. epidermidis may infection in the United States. J Arthroplasty 2012;27(8 Suppl):61–65.e1. maintain stable growth, whereas more virulent bacteria, such as Langvatn H, Lutro O, Dale H, Schrama JC, Hallan G, Espehaug B, et al. Bacterial and hema- fi S. aureus, maynotbesoreadilyabletoprosper.Importantly,relativenum- tological ndings in infected total hip arthroplasties in Norway assessment of 278 re- visions due to infection in the Norwegian Arthroplasty Register. Open Orthop J 2015; bers of co-pathogens in polymicrobial PJIs are not expected to be equal, 9:445–9. making it potentially challenging to differentiate contaminants from Lentino JR. Prosthetic joint infections: bane of orthopedists, challenge for infectious – pathogens but also meaning that some polymicrobial infections may pos- disease specialists. Clin Infect Dis 2003;36(9):1157 61. fi Maradit Kremers H, Larson DR, Crowson CS, Kremers WK, Washington RE, Steiner CA, sibly be misclassi ed as being monomicrobial. et al. Prevalence of total hip and knee replacement in the United States. J Bone Another finding in our study was the high rate of enterococci (36%) Joint Surg Am 2015;97(17):1386–97. in polymicrobial PJIs. A multicenter European study focused on PJI solely Osmon DR, Berbari EF, Berendt AR, Lew D, Zimmerli W, Steckelberg JM, et al. Executive sum- fi mary: diagnosis and management of prosthetic joint infection: clinical practice guide- caused by enterococci identi ed 107 cases (54%) of polymicrobial lines by the Infectious Diseases Society of America. Clin Infect Dis 2013;56(1):1–10. infections out of a sample of 203 patients (Tornero et al., 2014). The Otto M. Staphylococcus epidermidis—the 'accidental' pathogen. Nat Rev Microbiol 2009;7 study also found that polymicrobial PJI occurred earlier after implanta- (8):555–67. Piper KE, Jacobson MJ, Cofield RH, Sperling JW, Sanchez-Sotelo J, Osmon DR, et al. Micro- tion than did monomicrobial PJI (Tornero et al., 2014). Despite the fact biologic diagnosis of prosthetic shoulder infection by use of implant sonication. J Clin that our sample of enterococcal PJI was small (n = 25), we found Microbiol 2009;47(6):1878–84. comparable results, with a higher proportion of polymicrobial PJI Stewart EJ, Payne DE, Ma TM, VanEpps JS, Boles BR, Younger JG, et al. Effect of antimicro- bial and physical treatments on growth of multispecies staphylococcal biofilms. Appl among our Enterococcus-positive samples. In the prior study, enterococ- Environ Microbiol 2017;83(12):e03483-16. cal co-pathogens were CoNS in 37 cases, S. aureus in 24 cases, E. coli in 13 Tande AJ, Patel R. Prosthetic joint infection. Clin Microbiol Rev 2014;27(2):302–45. cases, Pseudomonas aeruginosa in 14 cases, Enterobacter cloacae in 4 Tornero E, Senneville E, Euba G, Petersdorf S, Rodriguez-Pardo D, Lakatos B, et al. Charac- cases, and other microorganisms in 15 cases (Tornero et al., 2014). teristics of prosthetic joint infections due to Enterococcus sp. and predictors of failure: a multi-national study. Clin Microbiol Infect 2014;20(11):1219–24. Our sample of enterococcal PJI cases was too small to highlight any Trampuz A, Piper KE, Jacobson MJ, Hanssen AD, Unni KK, Osmon DR, et al. Sonication of pattern of association with microorganisms, other than S. epidermidis. removed hip and knee prostheses for diagnosis of infection. N Engl J Med 2007;357 (7):654–63. Interestingly, the proportion of S. aureus, Gram-negative bacilli, and fi — fi Tyner H, Patel R. Propionibacterium acnes bio lm a sanctuary for Staphylococcus aureus? streptococci was not signi cantly different between monomicrobial and Anaerobe 2016;40:63–7. polymicrobial PJI. Vergidis P, Greenwood-Quaintance KE, Sanchez-Sotelo J, Morrey BF, Steinmann SP, Karau There are some limitations to our study. It was a retrospective study MJ, et al. Implant sonication for the diagnosis of prosthetic elbow infection. J Shoulder Elbow Surg 2011;20(8):1275–81. of nonconsecutive patients, and not all hardware extracted in order to Zimmerli W, Trampuz A, Ochsner PE. Prosthetic-joint infections. N Engl J Med 2004;351 treat a PJI at the Mayo Clinic is sent for sonication. Furthermore, the (16):1645–54.